Polymorphism and Identification of Metallothionein Isoforms by

To date, characterization of MT polymorphism by RP HPLC has been based on the ... Electrospray MS experiments were performed using a PE-SCIEX API 300 ...
0 downloads 0 Views 115KB Size
Download PDF
Anal. Chem. 1998, 70, 2536-2543

Polymorphism and Identification of Metallothionein Isoforms by Reversed-Phase HPLC with On-Line Ion-Spray Mass Spectrometric Detection Hubert Chassaigne and Ryszard Lobinski*

Laboratoire de Chimie Bio-inorganique et Environnement, CNRS EP132, He´ lioparc, 2, avenue du Pre´ sident Angot, 64000 Pau, France

Microbore reversed-phase HPLC with on-line ion-spray mass spectrometric detection is proposed for a study of polymorphism of rabbit liver metallothionein (MTRL) and its major isoforms MT-1 and MT-2. Separation conditions had been optimized until each chromatographic peak could be attributed to a single metalloprotein species, of which the molecular mass could be determined by ionspray MS. At the optimized conditions (elution with the gradient 5-8% of acetonitrile within 50 min using a 5 mM acetate buffer at pH 6), the chromatogram of MTRL showed five peaks, whereas those of MT-1 and MT-2 showed nine and seven different peaks, respectively. The on-line determination of the molecular masses ((1 Da) of the compounds eluted permits the unambiguous cataloguing and further referencing of putative and true MT subisoforms. The results obtained are compared with those obtained by direct ion-spray MS of the MT preparations at different pHs, with a goal to identify possible chromatographic artifacts. Metals (Cd, Cu) in the eluted complexes were studied by HPLC with on-line ICP MS detection. Metallothioneins (MTs) are a group of nonenzymatic, lowmolecular-mass (6-7 kDa) metal-binding proteins, which are characterized by their resistance to thermocoagulation and acid precipitation, by the presence of ∼60 nonaromatic amino acids and of a large number of cysteinyl residues, and by the absence of disulfide bonding.1,2 Metallothioneins, isolated from liver, kidney, and brain samples, seem to play an important role in homeostatic control, metabolism, and detoxification of a number (Zn, Cu, Cd, Hg) of trace metals.2,3 Certain MTs were shown to exist as isoforms that are the product of genetic polymorphism characteristic of MT genes in animals and humans; consequently, they draw attention to studies of metal-mediated gene expression mechanisms.4 This polymorphism occurs during the evolution of a species and consists of

the variation of the primary structure by the substitution of 1-15 amino acids. Hence, isoforms differ in amino acid composition other than cystein residues and, consequently, have different isoelectric points and different hydrophobicities.5 Isoforms with minor differences, such as one amino acid residue, were detected as subgroups of the two major isoforms and are termed subisoforms.5 Whereas sequences of the major isoforms have been decoded,6 the difficulties in the separation and identification of subisoforms are responsible for the lack of literature data regarding their identity. Classic methods for the separation of MT isoforms are based on the differences in the electrical charge of the molecule induced by the variation of the primary structure and include anionexchange chromatography and electrophoresis.7-9 The order of elution from an anion-exchange column has been the basis of the existing nomenclature, MT-1 and MT-2.10 The difference in the electrical charge induced by the substitution of one or several amino acid in the polypeptic chain is often not sufficient to affect the capacity factor on an anion-exchange column. Reversed-phase chromatography was shown to discriminate several peaks (termed subisoforms) within the MT-1 and MT-2 classes.11-14 To date, characterization of MT polymorphism by RP HPLC has been based on the retention time and the intensity of UV absorption, which has led to speculative, confusing, and often contradictory data, because it is virtually impossible to know what species was detected. Indeed, the use of the apparently identical analytical techniques and operating conditions (column, mobile phase) does not guarantee that similar results for samples of the same origin are obtained.11,12,14 The chromatograms not only show different morphologies but also differ in terms of the number of

* To whom correspondence should be addressed. Tel.: +33 5 59 80 68 85. Fax: + 33 5 59 80 12 92. E-mail: [email protected]. (1) Stillman, M. J. Coord. Chem. Rev. 1995, 144, 461-511. (2) Stillman, M. J., Shaw, C. F., Suzuki, K. T., Eds. Metallothioneins Synthesis, Structure and Properties of Metallothioneins, Phytochelatins and Metalthiolate Complexes; VCH: New York, 1992. (3) Riodan, J. F., Valee, B. L., Eds. Metallobiochemistry Part B. Metallothionein and Related Molecules, Methods in Enzymology; Academic Press: New York, 1991; Vol. 205.

(4) Hamer, D. H. In Metallothionein III: Biological Roles and Medical Implications; Suzuki, K. T., Imura, N., Kimura, M., Eds.; Birkhauser: Boston, 1993. (5) Suzuki, K. T. In Metallothioneins Synthesis, Structure and Properties of Metallothioneins, Phytochelatins and Metalthiolate Complexes; Stillman, M. J., Shaw, C. F., Suzuki, K. T., Eds.; VCH: New York, 1992. (6) Ka¨gi, J. H. R. In Metallothionein III: Biological Roles and Medical Implications; Suzuki, K. T., Imura, N., Kimura, M., Eds.; Birkhauser: Boston, 1993. (7) Sato, M.; Suzuki, K. T. Biomed. Res. Trace Elem. 1995, 6, 13-28. (8) Lobinski, R.; Chassaigne, H.; Szpunar, J. Talanta 1998, 46, 271-289. (9) Richards, M. P.; Beattie, J. H. J. Cap. Electrophor. 1994, 1, 196-207. (10) IUPAC. J. Biol. Chem. 1977, 252, 5939-5941. (11) Klauser, S.; Ka¨gi, J. H. R.; Wilson, K. J. Biochem. J. 1983, 209, 71-80. (12) Richards, M. P.; Steele, N. C. J. Chromatogr. 1987, 402, 243-256. (13) Richards, M. P. Methods Enzymol. 1991, 205, 217-238. (14) Bordin, G.; Cordeiro Raposo, F.; Rodriguez, A. R. Can. J. Chem. 1994, 72, 1238-1245.

2536 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

S0003-2700(98)00030-4 CCC: $15.00

© 1998 American Chemical Society Published on Web 05/21/1998

peaks observed. The likely reason for these differences is the presence of ghost peaks in the chromatograms, possibly coming from products of oxidation (i.e., dimers or polymerized products), from different conformations, or from different metal composition (metalloforms). The absence of standards of sufficient and documented purity makes the unambiguous identification of an MT species in an HPLC eluate impossible without its tedious isolation and off-line sequencing. A way to establish the identity of peaks in RP HPLC as isoforms is to run the chromatography in both acidic and neutral buffer systems,11,15,16 but the poorer efficiency of HPLC under acidic conditions may lead to coelution of isoforms and subisoforms.17 The unambiguous confirmation of the peak identity, which is necessary to enable the comparison of data obtained for different samples, by different techniques and different authors, can be obtained only by the determination of the molecular mass of each of the eluting compounds. An electrospray probe (and its pneumatically assisted version, ion-spray) has recently gained considerable popularity for the accurate on-line determination of molecular mass of proteins.18-20 Ironically, the previous applications of ESI MS to the characterization of metallothioneins targeted the complexed metal rather than the protein itself. It was shown that EI MS with pH control can provide analysis of metals in native and reconstituted metallothioneins, showing how many and what cations are incorporated per molecule of metallothionein.21,22 LeBlanc optimized HPLC of a Cd4Zn3MT preparation and proposed a postcolumn acidification of the effluent to release the MT-bound metals in order to enable their subsequent determination by ion-spray MS, operated in source collision-induced dissociation mode.23 In all these studies, the “pure” rabbit liver MT-2 was studied without any concerns for its polymorphism. The objective of this study was to evaluate reversed-phase HPLC with on-line ion-spray MS detection for the characterization of the polymorphism of metallothioneins in order to determine the number of potential isoforms present and to enable their unambiguous referencing by means of their molecular mass. EXPERIMENTAL SECTION Apparatus. HPLC was performed using an ABI 140C microbore syringe pump, an ABI model 112A injection module, and an ABI model 785A absorbance detector equipped with a microbore cell (Applied Biosystems, Foster City, CA). Electrospray MS experiments were performed using a PE-SCIEX API 300 ionspray triple-quadrupole mass spectrometer (Thornhill, ON, Canada). BioToolBox software was used for the calculation of molecular masses and deconvolution of protein mass spectra. The ICP mass (15) Hunziker, P. E.; Ka¨gi, J. H. R. Biochem. J. 1985, 231, 375. (16) Hunziker, P. E.; Ka¨gi, J. H. R. Experientia Suppl. 1987, 52, 257-264. (17) Lin, L. Y.; Lin, W. C.; Huang, P. C. Biochim. Biophys. Acta 1990, 1037, 248. (18) Mann, M.; Meng, C. K.; Fenn, J. B. Anal. Chem. 1989, 61, 1702. (19) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1990, 62, 882-889. (20) Hofstadler, S. A.; Bakhtiar, R.; Smith, R. D. J. Chem. Educ. 1996, 73, A82A89. (21) Yu, X. L.; Wojciechowski, M.; Fenselau, C. Anal. Chem. 1993, 65, 13551359. (22) High, K. A.; Methven, B. A.; McLaren, J. W.; Siu, K. W. M.; Wang, J.; Klaverkamp, J. F.; Blais, J. S. Fresenius’ J. Anal. Chem. 1995, 351, 393402. (23) LeBlanc, J. C. Y. J. Anal. At. Spectrom. 1997, 12, 525-530.

spectrometer used in this work was the ELAN 6000 (PE-SCIEX). The sample introduction systems used was a Microneb 2000 direct injection nebulizer (DIN) (CETAC, Omaha, NE). A PE 410 LC BIO pump was used to provide a makeup flow. Reagents and Standards. Methanol and acetonitrile (SigmaAldrich) was of LC grade. Water purified using a Milli-Q water purification system (Millipore, Bedford, MA) was used. Liquid nitrogen (99.996%) was evaporated in situ and supplied at 40 psi. The buffer solution was prepared by dissolving 5 mmol L-1 ammonium acetate in water (or in 50% methanol or acetonitrile) and adjusting the pH to 6.0 with acetic acid. Buffers were sparged with helium to remove dissolved oxygen and thus to attain a nonoxidizing environment. This is an important consideration in light of the observed susceptibility of MTs to oxidation during isolation.13 Rabbit liver metallothionein MT-2 (34H95161), MT-1 (94H9504), and MTRL (56H9500) preparations were purchased from SigmaAldrich (Saint Quentin Fallavier, France). The preparations contained 6.0% Cd, 0.6% Zn, 0.5% Cu for MT-1, 5.3% Cd, 0.7% Zn, 0.5% Cu for MT-2, and 5.9% Cd, 0.5% Zn, 0.8% Cu for MTRL, as found by an independent ICP MS analysis. Batch-to-batch variation can be significant; the preparations should be checked for purity and fully characterized prior to their use as standards.24 The stock MT solution (1 mg mL-1) was prepared by dissolving 1 mg of metallothionein in 1 mL of water. Working solutions were prepared by the dilution of the stock solution with water or buffer as required. The stock solution was kept in the refrigerator at 4 °C in the dark. HPLC Conditions. Separations were carried out using a Vydac C8 150 mm × 1 mm × 5 µm column. The pump flow was set at 40 µL min-1, which corresponded to a pressure of 4.5 MPa. The solutions were degassed by sparging with helium. Injection volume was 5 µL. Buffer A was 5 mM acetate buffer in water (pH 6.0), and buffer B was 5 mM acetate buffer in 50% CH3CNH2O (pH 6.0). Metallothionein complexes were eluted with a linear gradient from 10% B to 16% B within 50 min. Electrospray MS Conditions. An analyte solution (at 5 µL min-1 for the direct mode and at 40 µL min-1 for HPLC effluents) was introduced via a fused silica capillary (100 µm i.d.) that was inserted into the ion-spray needle held at 4400 V (ion-spray voltage). The orifice potential was set to 60 V. MS data for apoMT and Cd4-MT complexes in direct analysis were acquired by scanning Q1 over the m/z ranges of 800-1800 and 900-1900, respectively, while Q2 and Q3 were operated in the rf mode only. Mass spectra were acquired using a 0.5-Da step size and a dwell time of 5 ms, resulting in a typical scan time of about 3 s. For HPLC analysis, limited range mass spectra of 1250-1450 and 1600-1800 were acquired for Cd4-MT and Cd7-MT complexes using a 0.5-Da step size and a dwell time of 1 ms, resulting in a scan time of about 0.8 s. ICP MS Conditions. The rf source power used was 1.2 kW. The X-Y position of the ICP torch was adjusted first to maximize the signal for cadmium. Then, the position of the DIN capillary was optimized with the same objective. The fine-tuning was performed by the optimization of the nebulizer gas pressure (between 50 and 80 psi). The nebulizer makeup gas was optimized from the software of the ICP mass spectrometer (24) Van Beek, H.; Baars, A. J. J. Chromatogr. 1988, 442, 345-352.

Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2537

Figure 1. Sequences of rabbit liver metallothioneins reported in the literature.6 Roman numerals describe the position of Cd atoms in R and β clusters. Cadmium-cysteine bonds are indicated by a line between a cadmium atom (I-VII) and a cysteine (C).

between 0 and 300 mL min-1. The makeup liquid was set at 60 µL min-1 to give a total flow at the nebulizer of 100 µL min-1. An amount of 15 mL min-1 of oxygen was added to the nebulizer gas (0.95 L min-1) by means of a mass flow controller. RESULTS AND DISCUSSION Polymorphism of MTs depends on the biological species; therefore, for the sake of consistency of this study, the discussion of the methodology developed will be limited to rabbit liver MTs, which are by far the best described MTs in the literature. This choice has also been dictated by the commercial availability of preparations of rabbit liver MT (MTRL) isolated by size-exclusion chromatography (SEC) and of the isoforms MT-1 and MT-2 purified further by anion-exchange chromatography. Despite uncertain purity and identity of the individual MTs present, these commercial preparations are widely used as reference samples. To date, six rabbit liver metallothionein isoforms have been isolated and sequenced in the literature.25 Their sequences are summarized in Figure 1. The terminology used to identify MTs is based on the IUPAC-IUB Commission on Biochemical Nomenclature (CBN).10 The numerical identifier refers to the order of elution of the MT variants from anion-exchange chromatography. The letter identifiers are used to denote homology of the decoded amino acid sequence for same variant. Preparations abundant in cadmium and with minimum contents of Cu and Zn were chosen to reduce the number of species potentially present in the system studied. Cadmium atoms (denoted by Roman numerals) are ligated in two clusters (R and β), the forms being very stable leading to potentially stable Cd4 and Cd7 complexes.21 Isoforms MT-2a and 2c contain 62 amino acids whereas other forms contain 61 amino acids. Analysis of MT Preparations by Infusion ESI MS Under Demetallating Conditions. Conditions found by Yu et al. to generate electrospray signal from a purified apo-MT-2 (10 µM acetic acid) were found insufficient to demetalate the MT-2.21 Good results were obtained with the acidification with hydrochloric acid (25 mM in 30% methanol, pH 1.8). (25) Hunziker, P. E. In Methods in Enzymology; Riordan, J. F., Valee, B. L., Eds.; Academic Press: New York, 1991; Vol. 205, pp 421-426.

2538 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

Figure 2. Reconstructed ion-spray spectra of rabbit liver apometallothionein preparations. (a) MT-2; (b) MT-1; (c) MTRL. The spectra were obtained by the deconvolution of three charge state envelopes observed in a mass spectrum obtained by feeding an aqueous solution containing 30% methanol into the source at the rate of 5 µL min-1 at pH 1.8 (25 mM HCl).

Characterization of the MT-2 Preparation. The ion-spray mass spectrum of apo-MT-2 (not shown) contains three envelopes of ionization states (+5, +6, and +7), with four well-resolved m/z signals in each envelope. This pattern is similar to that obtained by High et al.22 but in contrast to Yu et al.,21 who observed only two signals associated with each charge state envelope. In the latter case, the MT-2 preparation used was earlier purified by RP HPLC under acidic conditions. The raw IS MS spectrum (signal intensity vs m/z) allowed us to reconstruct a mass spectrum of the MT-2 preparation, which is shown in Figure 2a. For the purpose of further discussion, the six peaks present have been denoted as MT-2R, β, γ, δ, , and ζ. The two latter peaks are revealed only after the reconstruction of the IS MS spectrum and are likely to be artifacts. The peak at 6229 Da in the reconstructed spectrum is an artifact of the numerical deconvolution of the ionization state +15 (m/z 1039.0) of superoxide dismutase (M ) 15 570 Da) present in the preparation. The calculated molecular mass for the major isoform (MT2β), 6125.5 ( 0.5 Da, is identical with that calculated on the basis of the published sequence of MT-2a, 6125.3 Da. Using a similar instrument, LeBlanc reported 6125 and 6127.3 ( 1.8 Da from various experiments,23 Yu et al. 6126.1 ( 0.9 Da,21 and High et al. 6125 ( 1 Da.22 There is also an agreement regarding another 30-Da heavier isoform (MT-2δ) among the authors: 6155.3 (calculated) and 6155.5 Da (this work), which is consistent with

the A/T heterogeneity of MT-2c reported by Hunziker25 (cf. Figure 1). Our work reveals the potential presence of at least four other species, clearly seen in the deconvoluted spectrum in Figure 2a. Of these previously not accounted for forms, one MT-2γ (M ) 6146.0 Da) corresponds perfectly to apo-MT-2b with the A/T, N/S, T/K, A/E, P/A, and A/T heterogenities and the calculated molecular mass of 6146.3 Da. It should be noted that this mass could also correspond to an impurity of MT-1, of which the only accounted for form has a calculated molecular weight of 6145.3. The presence of this impurity is unlikely, since the MT-2 preparation had been purified by anion exchange, which should have separated MT-2 from MT-1. Also, a peak at 6214.0 Da can be attributed to an isoform that was denoted by Kimura et al.26 as MT-2d (6215.4 Da calculated with respect to 6214 Da found). The MT-2R at 6084.5 ( 0.5 Da can be precisely calculated on the basis of the original spectrum (not shown) but disappears in the noise of the deconvoluted spectrum. On the other hand, the apo-MT2e (calculated molecular weight of 6241.5) could not be identified in the analyzed preparation. Forms MT-2R and MT-2 are not known in the literature and seem to be artifacts. This hypothesis is corroborated in the case of the form MT-2R by the discrepancy between the found 432.5-Da and theoretical 442-Da mass difference between the apo and Cd4 forms, whereas the form MT-2 is a mathematical artifact of the deconvolution of the m/z spectrum. The morphology of three major and two minor peaks looks similar to that reported by Bordin et al. by RP HPLC/UV, but this comparison may be accidental, since the response in ESI MS and UV of different isoforms may not be the same.14 Characterization of the MT-1 Preparation. Electrospray mass spectra of MT-1 have, to our knowledge, never been reported in the literature. The mass spectrum of the apo-MT allows one to distinguish seven well-resolved m/z signals reproduced in three charge envelopes. The ionization of MT-2 is apparently easier than that of apo-MT-1 because more intense signals are observed at the similar ionization conditions. The deconvoluted ion-spray mass spectrum (Figure 2b) shows the presence of five or six major peaks and an important number of minor peaks in the molecular weight range between 6100 and 6500. For the purpose of further discussion, the peaks have been denoted MT-1R-ζ, with the exception of the first signal in the +6 and +7 envelopes, which was identified as being identical with that belonging to the major apo-MT-2β (a likely impurity). The major peak (M ) 6144.0 Da) is likely to correspond to the only sequence reported in the literature for the MT-1 rabbit liver4 of 6145.3 Da. This morphology does not correlate with that observed by Bordin et al., who attributed two major and three minor isoforms for the MT-1 preparation.14 A protein with molecular mass of 8295 Da is observed in the MT-1 preparation at pH 4.0 and at higher pH. Characterization of the Rabbit Liver Metallothionein Preparation. The spectrum of the rabbit liver MT preparation (which had not undergone anion-exchange chromatography), reconstructed on the basis of the +4, +5, and +6 ionization states, is shown in Figure 2c. The spectrum is not a simple overlap of the MT-1 and MT-2 spectra because of an excess (4-fold14) of MT-2 (26) Kimura, M.; Otaki, N.; Imano, M. In Metallothionein; Ka¨gi, J. H. R., Nordberg, M., Eds.; Birkha¨user Verlag: Basel, 1979.

and its higher response in comparison with MT-1. The spectrum shows two envelopes of peaks in the ranges 6100-6250 and 63506500 Da. In the first envelope, three peaks corresponding to the MT-2a, MT-2b, and MT-2c isoforms are clearly seen. The 2 mass units of difference may introduce some ambiguity to this value, but the original ion-spray spectrum (not shown) shows clearly the presence of the +6 (1021.5 Da) and +5 (1226.0 Da), corresponding to the MT-2a. The spectrum corroborates the statement that the most abundant isoform in RL was a subspecies of MT-2.12 Peaks corresponding to the MT-1 isoform are less abundant; nevertheless, the most abundant forms, MT-1δ and MT1, can be identified. The second envelope lies ∼250 Da higher and contains wellresolved peaks MTRL R, β, γ, δ, which cannot be matched with the published sequences. These peaks are supported by peaks in the original spectra at m/z ratios corresponding to the +4 and +5 charge states in the range 1594-1624 and 1275.5-1299 Da, respectively. The further acidification to pH 1.2 (with HCl) does not make these peaks disappear; it is, therefore, not likely that they can be attributed to cadmiated MTs. The origin of this envelope is not clear, since it does not appear again in either of the spectra of purified MTs. Analysis of MT Complexes with Cadmium by Infusion ESI MS. Metallothioneins were shown to retain conformation in the gas phase that is similar to that in solution and not to fragment in an electrospray source.21-23 The cadmium content in the MT preparations analyzed is sufficient to obtain metallocomplexes at appropriate pH conditions. An analysis of the metalation of a MT as a function of pH allows one to detect the true metallothioneins and to discuss the possibility of electrospray artifacts. Deconvoluted mass spectra of MTRL, MT-1, and MT-2, obtained at pH 4 (5 mM acetate buffer in 30% CH3OH), in which the metalated R-domain should be present, are shown in Figure 3. A comparison of the apo-MT-2 and Cd4-MT-2 spectra confirms the 6229.0 Da seen in the apo-MT-2 spectrum (Figure 2a) to be an artifact. Regarding the other signals, the pattern of the Cd4MT spectrum follows that of the apo-MT spectrum. Some doubt may occur for MT-2ζ, for which the signal is not intense. Also, in the case of MT-1, the pattern of the mass spectrum of the Cd4 form (Figure 3b) follows that of the apo form. Except for one peak (6538.5 Da), which appears before the major envelope, the molecular masses of all the others correspond to signals observed in the spectrum of apo-MT-1. The comparison of the apo-MTRL and Cd4-MTRL spectra (Figure 3c) is less straightforward. In the first envelope, the form MT-2R appears in the Cd4 spectrum. Forms MT-2a and MT-2c appear to be enriched, with four atoms of Cd in the Cd4 spectrum. Form MT-2b disappears, maybe because it is not correctly resolved from the background. The latter is true also for the form MT-1 tentatively identified in the apo-MTRL spectrum. In the second envelope, the addition of four Cd atoms is clearly reflected by the molecular masses observed in the spectrum. One of the peaks (MTRLβ) is not present in the metalated spectrum and, thus, is likely to be an artifact. The remaining three peaks have the appearance of true isoforms. Further metalation (by gradual increase of pH) does not bring much additional information. At pH 5 (5 mM acetate buffer in 30% CH3OH), the metalated β-domain exists in equilibrium with the fully metalated form for all the preparations studied. The Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2539

Figure 3. Reconstructed ion-spray spectra of rabbit liver Cd4metallothionein complexes. (a) Cd4-MT-2; (b) Cd4-MT-1; (c) Cd4-MTRL. The spectra were obtained by the deconvolution of three charge state envelopes observed in a mass spectrum obtained by feeding an aqueous solution containing 30% methanol into the source at the rate of 5 µL min-1 at pH 4 (5 mM acetate buffer).

equilibrium is shifted toward the fully metalated MT at pH 6 and higher. Spectra at pH 6 and 7 show the shift of the envelope by three more cadmium atoms toward higher masses. The individual peaks, however, cannot be clearly seen. Characterization of the MT Polymorphism by Direct ESI MS. Table 1 summarizes the molecular masses determined for

all the peaks observed in the mass spectra of demetalated and metalated metallothioneins. It can be seen that, in most of the cases, the difference between the molecular mass is close to the theoretical value of 442 Da, according to the formula M(Cd4-MT) ) M(apo-MT) + 4Cd - 8H. The difference in some cases is larger, e.g., for MT-1β it is -8 Da; for MT-2R, -10 Da; while for MTRLδ it is +8.5 Da. The origin of this difference remains, for the moment, unknown. The data exclude the presence of other forms which are hidden in the noise of the spectrum. Some data in Table 1 can be compared with literature. For Cd4-MT-2a, the mass of 6566.5 ( 1.0 Da was obtained by High et al.22 and Yu et al.21 LeBlanc23 reports 6569.5 ( 0.8 Da (6568.9 ( 1.7 Da; 6569.7 ( 0.7 Da from other experiments) and 6570.3 ( 1.9 Da. For Cd7MT-2a, the values of 6901.3 ( 1.7 Da23 and 6897.6 + 0.9 Da (Cd7MT-2a with four extra hydrogen atoms, expected mass 6898.0 Da)23 were found. The slight discrepancies between the calculated and found values can be attributed to the different number of H• radicals associated with the complex.21 The data found in direct measurements allow the calculation of Cd7-MT masses (according to the formula M(Cd7-MT) ) M(apoMT) + 7Cd - 14H) which could not be determined by a direct measurement. These masses are going to be used further as reference values for mass spectra taken at the apex of each of the chromatographic peaks. Molecular masses calculated on the five of six published sequences can be attributed to peaks present in the spectra. Some molecular mass values (MT-1a and MT-2b) are too close, and a preseparation (by anion exchange) is needed to enable their unambiguous attribution. Reversed-Phase HPLC with On-Line ESI MS Detection. The ambiguity in the case of similar molecular masses of MT species can be avoided by separating the species and calculating the molecular masses on-line of the eluting compounds. Since the MT-1 and MT-2 preparations are pure in terms of SEC and anion-exchange chromatography, an orthogonal separation mechanism, reversed-phase chromatography, was optimized. A microbore column was chosen to ensure the compatibility of the separation flow rate with those tolerated by the electrospray

Table 1. Molecular Masses of Putative Rabbit Liver Metallothionein Isoforms Found by Direct Infusion Ion-Spray MS molecular mass (Da) compounds

apo-MT

Cd4-MT

Cd4-MT - apo-MT

Cd7-MT calcda

MT-2 R β γ δ  ζ

6084.5 6125.5 6146.0 6156.0 6180.0 6214.0

6517.0 6566.0 6585.0 6596.0 6626.0 6656.0

432.5 440.5 439.0 440.5 446.0 442.0

6857.8 6898.8 6918.8 6928.8 6952.8 6986.8

MT-1R β γ δ  ζ MTRL R β γ δ

6144.0 6176.0 6192.0 6215.0 6238.5 6272.0 6372.0 6420.0 6456.0 6493.0

6587.0 6610.0 6636.0 6657.0 6684.0 6714.0 6817.5

443.0 434.0 444.0 442.0 445.5 442.0 445.5

6898.0 6943.5

442.0 450.5

6916.8 6948.8 6964.8 6987.8 7012.8 7044.8 7144.8 7192.8 7228.8 7265.8

a

M(Cd7-MT) ) M(apo-MT) + 7Cd - 14H.

2540 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

M sequences apo-MT

nomenclature

6125.3 6146.3 6155.3

MT-2a MT-2b MT-2c

6215.4 6241.5 6145.3

MT-2d MT-2e MT-1a

Figure 4. Separation of MT-2 putative isoforms by microbore RP HPLC. A chromatogram obtained with spectrophotometric detection for 1.0 µg of MT-2 at pH 6.0. Gradient: 0-50 min, 5-8% CH3CN. The m/z and relative abundances of the peaks present in the ionspray mass spectrum taken at the apex of each of peaks 1-7 have been summarized in Table 2.

Figure 5. Separation of MT-1 putative isoforms by microbore RP HPLC. A chromatogram obtained with spectrophotometric detection for 1.0 µg of MT-1 at pH 6.0. Gradient: 0-50 min, 5-8% CH3CN. The m/z and relative abundances of the peaks present in the ionspray mass spectrum taken at the apex of each of peaks 1-9 have been summarized in Table 3.

source. Since the separations of apo-MT require the use of a 0.1% TFA (15 mM) buffer that suppresses the ESI signal, we decided to work with neutral buffers. Both the native peptide conformation and the original metal composition are preserved at neutral pH, retention times are shorter, and recoveries are high (90%), in contrast to those observed with acidic buffers.13 Methanol and acetonitrile, the most widely used mobile phases,13 were examined in detail. The choice between these solvents is a tradeoff between the efficiency of separations between the individual isoforms and the sensitivity of their detection. The advantage of methanol over acetonitrile is better ionization conditions of analytes in an electrospray source and better tolerance by the ICP. Methanol was reported to allow one to separate MT-2 and MT-1 on a C8 reversed-phase column with a linear gradient to 30% methanol with a neutral buffer.24 On the other hand, a series of initial experiments showed that replacing methanol by acetonitrile leads to an increase in the separation efficiency and an increase in the number of peaks observed. Therefore, acetonitrile was chosen for further studies with ESI MS detection. In the case of ICP MS, methanol was preferred. Characterization of the MT-2 Preparation. The chromatogram (Figure 4) shows one major peak preceded by four or five minor peaks and followed by one minor peak. Table 2, which summarizes the mass spectra taken at the apex of each chromatographic peak, shows the possibility of attributing a single

molecular mass to each of the peaks seen in the chromatogram. The mass spectra of peaks 2-5 shows single sharp peaks, which enable the determination of the molecular masses of the eluting compounds. Mass spectra of peaks 1 and 7 are noisy. The major isoform is considerably enriched with respect to the neighboring peak, corresponding to the R and γ species. Peak 5 (Cd7-MT-1a) is not well separated from the major MT-2a (peak 6) because of the disproportion in abundance of these two compounds. When similar concentrations of these species are present in the sample, a separation can be obtained (cf. peaks 7 and 8 in Figure 5). Table 2 shows that peaks with an identical molecular mass can elute at different retention times (peaks 1 and 5, M ) 6915.0 ( 0.5 Da). The complexity of the MT polymorphism makes simple ESI MS not sufficient. A two-dimensional analysis (retention time and mass spectrum) is necessary. Characterization of the MT-1 Preparation. For the MT-1 preparation, nine peaks of roughly similar intensities can be seen (Figure 5), in comparison with seven in the MeOH chromatogram (not shown). Most of the peaks contain one predominant compound well separated from minor impurities. Peaks 2 and 8 are the exceptions, where mass spectra taken at the peak apex show two peaks with similar relative abundances. Cd7-MT-1a (peak 7) and Cd7-MT-2a (peak 8) are better separated than in the case of the MT-2 preparation (cf. peaks 5 and 6 in Figure 4). Masses 6989 ( 1 and 6915.0 ( 0.5 Da (MT-1a) are resolved,

Table 2. Mass Spectrometric Data and the Relative Abundances of Peaks Present in a Mass Spectrum of Each of the Peaks Present in the TIC Chromatogram of MT-2 Preparation mass spectrum m/z values (amu) peak

[M + 5H]5+

[M + 4H]4+

relative abundance (%)

molecular mass (Da)

1 2 3 4 5 6

1384.0 1374.0 1386.0 1373.0 1384.0 1372.0 1381.0 1387.0 1387.0

1730.0 1717.0 1731.0 1715.0 1730.0 1716.0 1726.0 1734.0 1734.0

100 100 100 100 100 5 80 15 100

6915.0 ( 0.5 6867 ( 3 6920.0 6851 ( 1 6915.0 ( 0.5 6857 ( 2 6900.0 6930.0 6930.0

7

compound

Cd7-MT-2γ Cd7-MT-1R Cd7-MT-2R Cd7-MT-2β Cd7-MT-2δ Cd7-MT-2δ

Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2541

Table 3. Mass Spectrometric Data and the Relative Abundances of Peaks Present in a Mass Spectrum of Each of the Peaks Present in the TIC Chromatogram of MT-1 Preparation mass spectrum m/z values (amu) peak

[M + 5H]5+

[M + 4H]4+

relative abundance (%)

molecular mass (Da)

1 2

1352.0 1391.0

1689.0 1738.0 1722.0 1717.0 1722.0 1731.0 1734.0 1746.0 1737.0 1748.0 1718.0 1730.0 1726.0 1734.0 1754.0

100 40 60 10 90 100 10 90 10 90 10 90 60 40 100

6754 ( 1 6951 ( 1 6884.0 6864.0 6884.5 ( 0.5 6922.0 6931 ( 1 6980.0 ( 1 6944.0 6989 ( 1 6868.0 6915.0 ( 0.5 6900.0 6931 ( 1 7011 ( 1

3 4 5

1379.0 1386.0 1387.0 1397.0

6 1399.0 7 8 9

1384.0 1381.0 1388.0 1403.0

compound Cd7-MT-1β

Cd7-MT-2γ Cd7-MT-2δ Cd7-MT-1δ Cd7-MT-1R Cd7-MT-2β Cd7-MT-2δ Cd7-MT-1

Table 4. Mass Spectrometric Data and the Relative Abundances of Peaks Present in a Mass Spectrum of Each of the Peaks Present in the TIC Chromatogram of MTRL Preparation mass spectrum m/z values (amu) peak

[M + 5H]5+

[M + 4H]4+

relative abundance (%)

molecular mass (Da)

1

1374.0 1379.0 1389.0

1717.0 1725.0

2 3

1384.0 1373.0 1384.0 1390.0

60 10 25 5 100 5 80 10 5 5 80 15 10 60 10

6864.5 6893 ( 3 6940.0 6912.0 6919 ( 4 6862 ( 2 6915.0 ( 0.5 6944.0 ( 0.5 6972.0 6857 ( 2 6900.0 6936 ( 4 6931 ( 1 6962 ( 2 6995.0

4 5

1372.0 1381.0 1389.0 1387.0 1393.0 1400.0

1729.0 1732.0 1717.0 1730.0 1737.0 1744.0 1716.0 1726.0 1734.0 1734.0 1742.0

whereas, with methanol as mobile phase, compounds with these masses are contained in the same chromatographic peak (not shown). The increased resolution enables the identification of some peaks with precise molecular masses which have not been seen earlier. Table 3 summarizes the molecular masses of species identified in the peaks and attempts to correlate them with the results obtained earlier. The early-eluting compounds (peaks 1-3 in Figure 5, for which no matches were found, may be complexes with other metals, e.g., copper. Indeed, a chromatogram obtained by RP HPLC/ ICP MS (Figure 6) shows strong copper signals associated with the first three Cd-containing compounds. Some copper is also found associated with the late-eluting Cd-MTs. It was attempted to find out whether pure Cu-MT or mixed Cd-Cu-MT complexes are involved. For this purpose, the eluent from the column was acidified by adding formic acid via a T-piece to demetalate the eluting complexes. Successful demetalation was obtained for peaks 4-9, where the corresponding apo-MT forms (in some cases accompanied by the corresponding Cd4-MT forms) were observed in the mass spectra, confirming the positive identification of Cd7 complexes (Table 3). In the spectra of peaks 1-3, however, signals with 2542 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

compound

Cd7-MT-2γ Cd7-MT-1R Cd7-MT-2R Cd7-MT-2β Cd7-MT-2δ Cd7-MT-1γ

Figure 6. Separation of MT-1 putative isoforms by microbore RP HPLC with ICP MS detection at pH 6.0. Gradient: 0-50 min, 0-50% CH3OH; 50 ng of MT-1 injected; regular line, 114Cd, bold line, 65Cu.

masses corresponding to neither an apo form nor cadmiated species were found. Assuming that copper was present in the eluted compounds, molecular mass matches were found corresponding to the mixed Cd2Cu2-MT-2δ (6513 ( 1 Da), Cd2Cu-MT1R (6441 ( 1 Da), and CdCu4-MT-1β (6537 ( 1 Da) complexes.

Figure 7. Separation of MTRL putative isoforms by microbore RP HPLC. A chromatogram obtained with spectrophotometric detection for 1.0 µg of MTRL at pH 6.0. Gradient: 0-50 min, 0-30% CH3OH. The m/z and relative abundances of the peaks present in the ionspray mass spectrum taken at the apex of each of peaks 1-5 have been summarized in Table 4.

The ESI MS detection after postcolumn acidification did not allow for the identification of any Cu complex associated with Cu signals other than those discussed above, despite some trace copper signal associated with the last Cd peaks in the chromatogram with the ICP MS detection (Figure 6). Rabbit Liver Metallothionein. For unknown reasons, ionization of the MTRL preparation in an electrospray source is less efficient than that of the purified MT-1 and MT-2 preparations. The use of acetronitrile as the mobile phase leads to a relatively noisy baseline in mass spectra taken at the apex of the chromatographic peaks, leading to ambiguities in the determination of the molecular masses. Since, in contrast to the purified preparations, the separation obtained with methanol was not found to be less efficient than that obtained with acetonitrile, the former solvent was used in the mobile phase. A chromatogram of the MT rabbit liver preparation which had not been purified by anion-exchange is shown in Figure 7. The chromatogram shows three major and two minor peaks. The mass spectrometric data of each of the peaks are summarized in Table 4. The chromatographic peaks lack the mass spectrometric purity; nevertheless, the resolution of the mass spectra registered on-line allows the precise determination of the molecular mass of the coeluting compounds.

The MS spectra taken at the apex of each of the chromatographic peaks allow one to interpret clearly the chromatogram in Figure 7 as that of a mixture of MT-1 and MT-2. Peaks 2 and 4 correspond to the major isoforms MT-2 (2γ and 2β, respectively), whereas peaks 3 and 5 corrrespond to the MT-1 isoforms (1R and 1γ). The analysis of the MS snapshots of peaks 3 and 4 confirms the elution of MT-1a before MT-2a. The form MT-1γ elutes after. The chromatogram contains a peak1 which does not appear in MT-1 and MT-2 preparations purified by anion exchange. Also, several peaks present in the purified preparations are not present in the MTRL chromatogram. Note also that the forms MTRLR-δ (the high-molecular mass envelope of the infusion ESI MS spectrum) are observed neither in TIC nor in UV chromatograms. It is, therefore, probable that these species are retained on the column or elute with the dead volume. CONCLUSIONS Reversed-phase HPLC with on-line ion-spray mass spectrometric detection is a powerful and promising tool for the characterization of the complex genetic polymorphism of mammalian metallothioneins. The combination of the resolving capability able to discriminate between MTs that vary by one amino acid only with the possibility of the instantaneous, very precise on-line determination of molecular mass opens a way to the identification and cataloguing of the naturally occurring MT isoforms. This technique should enable the verification of the often speculative and contradictory conclusions drawn on the basis of chromatography with UV detection and shed new light on the problem of chromatographic peak purity and the formation of artifacts. The molecular mass, which can now be readily accessed, is a potentially interesting parameter to systematize the nomenclature of MT isoforms. The HPLC and, probably, CZE data shown in the literature must be watched carefully when assigning the term “isoform” to a peak observed in a chromatogram.

Received for review January 7, 1998. Accepted March 25, 1998. AC9800309

Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2543